Real-Time Camouflage Gait Detection Model Based On Apache Spark Streaming
Ye Yan, Li PeiGuo · 2020
A real-time camouflage gait detection model is proposed. The model uses model-based feature representation. For mitigating the variation of the gait cycles, DTW Barycenter Averaging (DBA) is introduced to compute the average feature of the successive gait cycles. Using the Dynamic Time Warping (DTW), the model is capable of camouflage gait detection. Apache Flume and Spark Streaming are used for real-time processing. The model is evaluated using two Kinect gait datasets. Results showed that the proposed model is effective for real-time gait sequences processing, and it has an acceptable accuracy of camouflage detection.